Keir Starmer, the Prime Minister, made a speech last week at UCL East. The ‘typescript‘ on the Government website is astonishing. The whole thing is printed in the style of a Mr Men book. Now, perhaps this is justified by the fact that Starmer was speaking rather than writing: but it still seems to be somewhat alarming that politicians do not speak in sentences but in a sort of liturgical dirge style in which there is a continual doleful optimism.
Contemplate the Mr Men style in the opening lines of the speech:
Deb Kelly is a prison officer and a PE instructor.
Two years ago, this month…
She got up on a Saturday morning…
And collapsed on the floor.
Her whole face had completely drooped.
Her left side had gone completely weak.
She was having a stroke.
She was found by her son, rushed to hospital…
Where the doctors used Artificial Intelligence…
To help pinpoint the exact location of the blood clot.
They successfully removed it.
“Well done!” said Mr First Lord of the Treasury.
This is odd. It is paginated as if it is poetry. Perhaps Starmer had it written like this so he could offer it to the London Review of Books as a poem. Or so the Roger Hargreaves Estate could add it to the canon.
That’s the form.
The content is AI.
Starmer is keen on AI. He thinks it will “transform” things. Three things. He uses the word “transform” three times. AI will transform “the lives of working people”. It will transform “our entire understanding of biology”. It will transform “our public services”.
Biopower!
Anyhow, Starmer is keen on AI, and keen to reassure us 1. that there is no serious danger, 2. that there is a Government agency tasked with reassuring us that there is no danger. How? By regulating AI. Good.
What is AI?
Answer 1. It is a grandiose assimilator of words, also replicator, digestor: a self-teaching, massively cancerous self-teaching computer programming system. For some reason, everyone seems to accept it as an inevitability. Why?
Answer 2. Why? Well, it is the solution to All Our Problems. It will replace Theresa May’s Magic Money Tree, it will plaster over Rachel Reeves’s Black Hole, it will break strikes for the stay-at-home British Work Ethic, and it will compensate for our lack of Common Sense. It will write every application letter, compose every undergraduate essay, and write every political speech from now on.
Here is Starmer. AI will (and I quote) “help in the fight against tax avoidance… halve the time social workers spend on paperwork… make public services more human… [give doctors and nurses] more time for the personal touch… turbocharge every single element of our Plan for Change”.
This is funny and scary. AI will be so busy doing what a doctor should do that the person formerly called ‘Doctor’ will now be called a ‘Bedside Mannerer First Class’ while the person formerly called ‘Nurse’ will now be called a ‘Beside Mannerer Second Class’. And notice how much of this AI is to be used to ensure that the Kraken Government maintains its grip on the Ship of State: by fighting tax avoidance and turbocharging plans for change.
If the benefits of AI are the first part of his speech, the impetus to become a leader in AI technology is the second part. “Britain is going to shape the future.” “This is the nation of Babbage, Lovelace and Turing,” he says, listing three figures as little like Starmer as it is possible to be. He then lists a lot of companies I have never heard of, which are clustering around, attracting and extracting much cash – OpenAI, Anthropic, Scale, Mistral AI, Wayve, Synthesia, Blackstone, Kyndryl, Nscale, Vantage Data Centres – and likens what is happening now to the industrial revolution. He says that Britain was once “the cradle of engineering innovation”, and presumably hopes that it will be the nursing home of engineering innovation too.
The bit that puzzled me most was this section of the speech:
And, then of course, the engine of AI progress…
…is what’s called compute.
We’ll increase our public sector compute…
Not by a factor of two or three or even 10….
But by 20.
Eh? Is he using “compute” as a noun? My dictionary knows no such thing. I asked Google, and I found a discussion thread on Reddit from 2020 where someone notes that “compute” has become a noun, and someone else explains that it is common language in “cloud-related” or “cloud-adjacent” industries. Perhaps people don’t want to use “computation” any more. I wonder if the emphasis is on the first syllable. (ComPUTE is the verb; COMpute is the noun, perhaps?) I check the video of the speech on YouTube (at around 12 minutes in) and find that Starmer is a bit unsure: first, he says, “it’s what’s called comPUTE…”, then he adds, “We’ll increase our public sector COMpute…”)
This has to be translated into English.
The English translation is:
I, Keir Starmer, propose that the Government magnifies its power by employing AI knowledge-assimilating power-grabbing linguistic tools.
(Knowledge is power, quoth Francis Bacon, four centuries ago.) He is a bit clearer in the text of the article he published in the Financial Times on January 13th to coincide with his speech.
Britain should be excited by this. For one, it offers credible hope of a long-desired boost in public sector productivity. Nurses, social workers, teachers, police officers — for millions of frontline workers, AI can give the precious gift of time. This means they can refocus on the care and connection aspects of their job that so often get buried beneath the bureaucracy. That’s the wonderful irony of AI in the public sector. It provides an opportunity to make services feel more human.
“Public sector productivity” is as bad as “public sector compute”. (Though I suspect it may mean that frontline workers will have more time for the shirking parts of their jobs.) But this “public sector compute” seems to be a euphemism for what is elsewhere called “sovereign AI”. CNBC Africa comments:
Sovereign AI has become a hot topic for policymakers, particularly in Europe. The term refers to the idea that technologies critical to economic growth and national security should be built and developed in the countries people are adopting them in.
The UK wants a Little England (or Chinatown) version of OpenAI’s ChatGPT so that it can carry out more coups d’état against its poor subjects. And of course the private corporations want a piece of the action. One CEO (of a company called Antler), a man with the alarming name of Magnus Grimeland, identified the £7 trillion pounds of English pension funds as a “pocket” – his word – that he would like to pick. “Imagine if you take just 5% of that and allocate it to innovation — you solve the problem.”
Starmer himself in the FT adds: “AI has arrived as the ultimate force for change and national renewal.”
Good Lord. Is anyone thinking this through?
What I worry about is this. Many significant grim commentators are saying that the United Kingdom – even if we ignore Brexit, Covid, Climate, Diversity and Immigration – has gone through colossal changes, economic on the one hand (Bank of England, OBR, excessive borrowing, bailing out banks, inflation, no joined-up policy etc.) and political and constitutional on the other (undermining Parliament by Devolution, the Supreme Court, Quangos etc.) If this is so, as it is, then isn’t Starmer hoping that AI will bail out this bloated superstate? He seems to be wagering not on putting the house in order, but in hoping for a miracle to come along, a “game changer” (and add other fashionable phrases, as Starmer does). In other words, he is what the haughty ladies of the Guardian call a ‘tech bro’.
It seems to me that there is a moral case against AI. Most of the things promised seem strictly unnecessary. And every time we turn to AI to do something we only demonstrate our unwillingness to do it ourselves. In relation to manual labour, technical advance was mostly a blessing: relieving us or our animals of hours of milling, reaping, cutting etc. But since AI is a linguistic system, what we are facing is the destruction or demoralisation of our own linguistic abilities. Why learn to write when AI can write for us?
Dr. James Alexander is a Professor in the Department of Political Science at Bilkent University in Turkey.
Editor’s note: Apart from anything else, AI keeps getting even simple things wrong. Twice this week the Google ‘AI Overview’ that now appears at the top of searches has given false information. In looking up a George Orwell quote, the AI Overview wrongly asserted that it came from 1984, when in fact it comes from a preface to Animal Farm.

The AI Overview also claimed, in a different search, that the correct phrase is “hone in on”, while “home in on” is a “common mistake”. In fact, as the top search result below the AI Overview, an article from Merriam-Webster, relays, the original phrase is “home in on”, derived from homing pigeons, and “hone in on” is a more recent corruption.

If AI can’t even get simple facts like these correct – facts confirmed in Google’s own top search results – how can it possibly be trusted with anything of any consequence?
To join in with the discussion please make a donation to The Daily Sceptic.
Profanity and abuse will be removed and may lead to a permanent ban.